illusory correlation n.
An apparent correlation that does not actually exist in the data being judged. In the classic demonstration of the illusion in 1967, the US psychologists Loren J(ames) Chapman (born 1927) and Jean Chapman (born 1929) presented experienced clinicians and students with information about a number of mental patients—for each patient, diagnostic statements and a drawing of a person made by that patient. The clinicians and students then estimated, from memory, the frequency with which different diagnostic statements (such as The man who drew this is suspicious of other people) had been associated with specific characteristics of the drawings (such as peculiar eyes in the drawing). Both groups of judges vastly overestimated the co-occurrences that fitted in with their implicit personality theories, and this illusion was extremely resistant to change, persisting even when the actual correlations were negative. For example, 91 per cent of the experienced clinicians and 58 per cent of the students reported that peculiar eyes had been associated with suspiciousness, and 80 per cent of the clinicians and 76 per cent of the students reported that broad shoulders had been associated with being worried about manliness. The illusion helps to underpin superstitions and prejudices, and it is usually explained by the availability heuristic, the assumption being that co-occurrences that seem likely are easier to remember than unlikely ones.